The Astrophysical Journal, 636:777–797, 2006 January 10 # 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.
THE H.E.S.S. SURVEY OF THE INNER GALAXY IN VERY HIGH ENERGY GAMMA RAYS F. Aharonian,1,2 A. G. Akhperjanian,3 A. R. Bazer-Bachi,4 M. Beilicke,5 W. Benbow,2 D. Berge,2 K. Bernlo¨hr,2, 6 C. Boisson,7 O. Bolz,2 V. Borrel,4 I. Braun,2 F. Breitling,6 A. M. Brown,8 P. M. Chadwick,8 L.-M. Chounet,9 R. Cornils,5 L. Costamante,2, 10 B. Degrange,9 H. J. Dickinson,8 A. Djannati-AtaI¨¨,ı 11 L. O’C. Drury,12 G. Dubus,9 D. Emmanoulopoulos,13 P. Espigat,11 F. Feinstein,14 G. Fontaine,9 Y. Fuchs,15 S. Funk,1, 2 Y. A. Gallant,14 B. Giebels,9 S. Gillessen,2 J. F. Glicenstein,16 P. Goret,16 C. Hadjichristidis,8 M. Hauser,13 G. Heinzelmann,5 G. Henri,15 G. Hermann,2 J. A. Hinton,2 W. Hofmann,2 M. Holleran,17 D. Horns,2 A. Jacholkowska,14 O. C. de Jager,17 B. Khe´lifi,2 Nu. Komin,6 A. Konopelko,2, 6 I. J. Latham,8 R. Le Gallou,8 A. Lemie`re,11 M. Lemoine-Goumard,9 N. Leroy,9 T. Lohse,6 J. M. Martin,7 O. Martineau-Huynh,18 A. Marcowith,4 C. Masterson,2,10 T. J. L. McComb,8 M. de Naurois,18 S. J. Nolan,8 A. Noutsos,8 K. J. Orford,8 J. L. Osborne,8 M. Ouchrif,10, 18 M. Panter,2 G. Pelletier,15 S. Pita,11 G. Pu¨hlhofer,2, 13 M. Punch,11 B. C. Raubenheimer,17 M. Raue,5 J. Raux,18 S. M. Rayner,8 A. Reimer,19 O. Reimer,19 J. Ripken,5 L. Rob,20 L. Rolland,18 G. Rowell,2 V. Sahakian,3 L. Sauge´,15 S. Schlenker,6 R. Schlickeiser,19 C. Schuster,19 U. Schwanke,6 M. Siewert,19 H. Sol,7 D. Spangler,8 R. Steenkamp,21 C. Stegmann,6 J.-P. Tavernet,18 R. Terrier,11 C. G. The´oret,11 M. Tluczykont,9, 10 G. Vasileiadis,14 C. Venter,17 P. Vincent,18 H. J. Vo¨lk,2 and S. J. Wagner13 Received 2005 June 10; accepted 2005 September 5
ABSTRACT We report on a survey of the inner part of the Galactic plane in very high energy gamma rays with the H.E.S.S. Cerenkov telescope system. The Galactic plane between 30 in longitude and 3 in latitude relative to the Galactic center was observed in 500 pointings for a total of 230 hr, reaching an average flux sensitivity of 2% of the Crab Nebula at energies above 200 GeV. Fourteen previously unknown sources were detected at a significance level greater than 4 after accounting for all trials involved in the search. Initial results on the eight most significant of these sources were already reported elsewhere (Aharonian and coworkers). Here we present detailed spectral and morphological information for all the new sources, along with a discussion on possible counterparts in other wavelength bands. The distribution in Galactic latitude of the detected sources appears to be consistent with a scale height in the Galactic disk for the parent population smaller than 100 pc, consistent with expectations for supernova remnants and /or pulsar wind nebulae. Subject headingg s: catalogs — gamma rays: observations — pulsars: general — supernova remnants — supernovae: general — surveys
Compton scattering on background radiation fields, whereas protons (and atomic nuclei) can generate gamma rays via the decay of neutral pions produced in hadronic interactions with ambient material. VHE gamma radiation is therefore expected from the acceleration sites of particles with energies above 1 TeV (1012 eV ). In particular, gamma rays can be seen as a probe of sources of nucleonic cosmic rays in our Galaxy and may thus provide part of the solution to the century old puzzle of the origin of this radiation.
1. INTRODUCTION 11
Very high energy (VHE, defined here as E > 10 eV ) gamma rays provide the most direct view currently available into extreme environments in the local universe. As gamma rays are produced by interactions of relativistic particles, they represent a good probe of nonthermal astrophysical processes. Relativistic electrons can produce gamma rays by nonthermal bremsstrahlung or by inverse 1 Correspondence and request for material should be sent to S. Funk, stefan
[email protected]. 2 Max-Planck-Institut fu¨r Kernphysik, Saupfercheckweg 1, Postfach 10 39 80, D-69029 Heidelberg, Germany. 3 Yerevan Physical Institute, Alikhanian Broothers Street, 375036 Yerevan, Armenia. 4 Centre d’Etude Spatiale des Rayonnements, CNRS/ UPS, Toulouse, France. 5 Universita¨t Hamburg, Institut fu¨r Experimentalphysik, Luruper Chaussee 149, D-22761 Hamburg, Germany. 6 Institut f u¨r Physik, Humboldt-Universita¨t zu Berlin, D-10115 Berlin, Germany. 7 Laboratoire de l’Univers et de ses theories, UMR 8102 du CNRS, Observatoire de Paris, Section de Meudon, France. 8 University of Durham, Department of Physics, South Road, Durham DH1 3LE, UK. 9 Laboratoire Leprince-Ringuet, IN2P3/CNRS, Ecole Polytechnique, Palaiseau, France. 10 European Associated Laboratory for Gamma-Ray Astronomy, jointly supported by CNRS and MPG.
11
Physique Corpusculaire et Cosmologie, Colle`ge de France, Paris, France. School of Cosmic Physics, Dublin Institute for Advanced Studies, 5 Merrion Square, Dublin 2, Ireland. 13 Landessternwarte Heidelberg, Ko¨nigstuhl, Heidelberg, Germany. 14 Laboratoire de Physique The´orique et Astroparticules, IN2P3/CNRS, Universite´ Montpellier II, France. 15 Laboratoire d’Astrophysique de Grenoble, INSU/CNRS, Universite´ Joseph Fourier, France. 16 DAPNIA / DSM /CEA, CE Saclay, Gif-sur-Yvette, France. 17 Unit for Space Physics, North-West University, Potchefstroom, South Africa. 18 Laboratoire de Physique Nucle´aire et de Hautes Energies, Universite´s Paris VI & VII, France. 19 Institut fu¨r Theoretische Physik, Lehrstuhl IV, Ruhr-Universita¨t Bochum, Germany. 20 Institute of Particle and Nuclear Physics, Charles University, Prague, Czech Republic. 21 University of Namibia, P.O. Box 86628, Windhoek, Namibia. 12
777
778
AHARONIAN ET AL.
Observations of synchrotron emission in the X-ray band provide strong indirect evidence for the existence of 100 TeVelectrons in shell-type supernova remnants (SNRs; see, e.g., Koyama et al. 1995) and in pulsar wind nebulae (PWNs; see, e.g., Weisskopf et al. 2000). At TeVenergies, shell-type SNRs and pulsars are established to be sources of VHE gamma rays as shown by highly significant detections of the Crab Nebula by the Whipple collaboration (Weekes et al. 1989) or the detection of the SNR RX J1713.73946 by the CANGAROO (Enomoto et al. 2002) and the High Energy Stereoscopic System (H.E.S.S.) collaborations (Aharonian et al. 2004c). However, despite the fact that SNRs are the best candidates for the acceleration of hadronic cosmic rays up to the so-called knee in the cosmic-ray particle spectrum at 1015 eV, there is as yet no generally accepted evidence for the existence of energetic nuclei (as opposed to electrons) in these objects. SNRs are favored as the site of cosmic-ray acceleration for two principal reasons: (1) a theoretically well-established acceleration mechanism exists in the form of diffusive shock acceleration at SNR shock fronts (for reviews see, e.g., Blandford & Eichler 1987; Malkov & Drury 2001), and (2) the Galactic supernovae and the resulting SNRs are the only known potential sources that can provide the necessary amount of energy (Ginzburg & Syrovatskii 1964). To explain the observed energy density of cosmic rays, supernova explosions must convert kinetic energy to energy of relativistic particles with a conversion efficiency of roughly 0.1–0.2. Theoretical expectations for the gamma-ray emission of SNRs arising from hadronic interactions are given by Drury et al. (1994) as F (>E ) 9 ; 1011
E 1 TeV
1:1
S cm2 s1 ;
ð1Þ
with S¼
ESN 1051 ergs
d 1 kpc
2
n ; 1 cm3
ð2Þ
where d denotes the distance to the object, n is the density of the circumstellar medium, assumed to be uniform, and ESN is the total kinetic energy released by the supernova. This formula assumes a power-law distribution of cosmic-ray energies with a typical differential photon index of 2.1. Calculations of the photon index give a range of values around the index value 2.0 for a uniform circumstellar medium (Berezhko & Vo¨lk 1997) and a rather more extended range of photon index values for supernova explosions into the wind bubble of a very massive progenitor star ( Berezhko & Vo¨lk 2000). A typical value for the fraction of kinetic energy converted into cosmic rays is ESN ¼ 1:0 ; 1050 ergs ( Malkov & Drury 2001; Vo¨lk et al. 2003). For an SNR at a distance of d ¼ 1 kpc expanding into an environment with a typical density of the interstellar medium of n ¼ 1 hydrogen atom cm3, a VHE gamma-ray flux of 20% of that from the Crab Nebula is expected. The luminosity L0 of such a source is 7:2 ; 1033 ergs s1 in the energy range between 0.2 and 10 TeV. One would therefore expect that with the sensitivity level of the H.E.S.S. Galactic plane survey (down to 2% of the Crab flux) SNRs should be detectable out to a distance of 3 kpc (to be compared to 8.5 kpc to the center of the Galaxy). For larger values of n, SNRs should be visible within most of the volume of the Galactic disk. Other potential sources of VHE gamma rays are pulsars and PWNs. Pulsars, rapidly rotating neutron stars left over after a
Vol. 636
supernova explosion, are highly magnetized and act like unipolar inductors accelerating particles. The pulsar wind interacts with the ambient medium, generating a shock region where particles are accelerated. Such objects may therefore exhibit a pulsed component of radiation, from the immediate vicinity of the pulsar, together with an unpulsed component from the shock region and beyond. The Crab Nebula is the best-studied object of this class of particle accelerators. Most of these potential Galactic gamma-ray sources such as SNRs and pulsars are associated with the formation of massive stars and therefore cluster along the Galactic plane, especially concentrated in the direction toward the center of our Galaxy. A total of 91 SNRs and 389 pulsars are cataloged within the inner 60 in Galactic longitude and 6 in Galactic latitude (Green 2004; Manchester et al. 2005). Despite the large number of known SNRs and pulsars in our Galaxy, the actual number of detected VHE gamma-ray sources was until now limited to only a handful of objects (for a review see, e.g., Weekes 2000). Particle acceleration to multi-TeVenergies may also be powered by OB stellar associations, via the superbubble mechanism (for a summary see Parizot et al. 2004) and also possibly more directly via the stellar winds of member stars and subsequent interaction (see, e.g., Montmerle 1979; Vo¨lk & Forman 1982; White 1985; Benaglia et al. 2001). There are presently two TeV sources that may be linked to such OB associations. TeV J2032+4130, discovered by the HEGRA collaboration (Aharonian et al. 2002a), and HESS J1303631, a recent discovery by H.E.S.S. (Aharonian et al. 2005d), are in the vicinity of the OB associations Cygnus OB2 and Centaurus OB1, respectively. Both TeV sources remain unidentified (UID) in the sense that there are as yet no clear counterparts of similar size and morphology in other wave bands. These sources might represent a new class of objects distributed along the Galactic plane. At lower energies, observations by the space-based instrument EGRET in the GeV energy range (Hartman et al. 1999) also revealed a population of sources clustering along the Galactic plane. The limited angular resolution of the EGRET instrument renders an identification of these sources in other wave bands very difficult, and most of these objects remain as yet unidentified. For all these reasons, a survey of the inner part of the Galaxy provides an efficient way to search for gamma-ray emission in order to investigate properties of source classes and search for yet unknown types of Galactic VHE gamma-ray emitters. At TeV energies the Milagro water Cerenkov detector (Atkins et al. 2004) and the Tibet air-shower array (Amenomori et al. 2002) have been used to perform large-scale surveys. While these instruments have the advantage of a very wide field of view (1 sr), the sensitivity obtained by these surveys is rather limited, reaching a flux limit comparable to the flux of the Crab Nebula, 3 ; 1010 cm2 s1 (for E > 200 GeV ), in 1 yr of observations. Both surveys covered 2 sr of the northern sky and revealed no previously unknown significant gamma-ray sources. However, some evidence for gamma-ray emission was found by the Milagro collaboration from a region in the Cygnus constellation and another region close to the Crab Nebula (Smith et al. 2005). Recently the detection of diffuse emission from the Galactic plane with Milagro has been reported ( Fleysher et al. 2005). The HEGRA instrument was the first array of imaging air Cerenkov telescopes to be used to survey a part of the Galactic plane (Aharonian et al. 2002b). The range of Galactic longitudes 2 < l < 85 was observed; due to the location of HEGRA in the Northern Hemisphere and the resulting large zenith angles for observations of the center of the Galactic plane, the sensitivity was reduced for the central part of the Galaxy and the energy
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
779
Fig. 1.— Off-axis performance of the H.E.S.S. instrument. Left: Gamma-ray efficiency relative to that for an on-axis source for Monte Carlo simulations (dashed line: 20 zenith angle; solid line: 45 zenith angle) and for H.E.S.S. observations of the Crab Nebula at a mean zenith angle of 47 (triangles). The Monte Carlo simulations are generated assuming a Crab-like source spectrum of the form dN /dE / E with ¼ 2:5. Right: Simulated angular resolution (68% containment radius) as a function of the offset for observations at zenith angles of 20 (circles) and 50 (triangles). In each plot the arrow denotes the offset of 2 up to which gamma rays are included in this analysis.
threshold of the observations ranged from 500 GeV to 7 TeV. No sources of VHE gamma rays were found in this survey, and upper limits between 15% of the Crab flux for Galactic longitudes l > 30 and more than 30% of the Crab flux in the inner part of the Milky Way were derived. Until the completion of H.E.S.S. (Hinton 2004) in early 2004, no VHE gamma-ray survey of comparable sensitivity of the southern sky or of the central region of the Galaxy had been performed. In this paper we describe a survey of this region with H.E.S.S. between 2004 May and July at a flux sensitivity down to 2% of the Crab flux. The initial results from the survey including the description of the discovery of eight new sources have been summarized in Aharonian et al. (2005f ). This work presents detailed results on these eight sources. In addition, we report the detection of six additional weaker sources of VHE gamma rays extending to lower flux levels than our previous analysis. Spectral information and a more detailed study on the morphologies for all sources are provided. The paper is structured as follows: After a description of the H.E.S.S. instrument (x 2) and the Galactic survey data set used in this paper (x 3), the data analysis technique is presented in x 4. In x 5 we give a detailed description of the individual sources along with possible counterparts in other wave bands, followed by a discussion of the results in x 6 and some conclusions in x 7. 2. THE H.E.S.S. INSTRUMENT H.E.S.S. is a system offour imaging atmospheric Cerenkov telescopes, located in the Khomas Highlands of Namibia at a height of 1800 m above sea level. H.E.S.S. has unprecedented sensitivity in the energy range from 100 GeV to several tens of TeV. The imaging Cerenkov technique exploits the cascade of secondary particles produced in interactions of primary gamma rays with molecules in the Earth’s upper atmosphere. These secondary particles emit Cerenkov light, which can be imaged by the Cerenkov telescopes. The energy and direction of the primary gamma rays can be reconstructed based on the intensity and shape of the detected Cerenkov light distribution. Multiple views of the same particle shower with several telescopes improve background rejection capability and the angular and energy resolution of the instrument, in comparison to a sin-
gle telescope. Therefore, H.E.S.S. observations are performed in a stereoscopic mode, whereby at least two of the four telescopes must be triggered by a shower for an event to be recorded (Funk et al. 2004). The telescopes each have a mirror area of 107 m2 ( Bernlo¨hr et al. 2003; Cornils et al. 2003). The Cerenkov light is focused onto fast 960 pixel photomultiplier cameras, which record the Cerenkov light images (Vincent et al. 2003). The 5 diameter field of view of the H.E.S.S. cameras is the largest of all imaging Cerenkov detectors and is especially suited for the study of extended sources and the search for unknown sources in surveys. This was part of the original design concept for the H.E.S.S. telescope system (Aharonian et al. 1997b, 1997c). The H.E.S.S. instrument has an energy threshold of 100 GeV at zenith before selection cuts, an angular resolution of 0N1 for individual gamma rays, and a point source sensitivity of 200 GeV ) and assuming a power-law spectrum with photon index ¼ 2:5. The sensitivity used here is defined as the minimum flux of a point source that would be detectable in this survey by H.E.S.S. at the 6.3 level (corresponding to 4 after accounting for all trials), given the cuts used to detect slightly extended sources (in particular a 2 ¼ 0:05 deg2 cut). The solid line shows the sensitivity averaged over the whole longitude range of the survey, including runs on the Galactic center region and on the SNR RX J1713.73946. The dashed line shows the sensitivity in the range of longitude 350 < l < 357 , a region where no reobservations were conducted.
As can be seen from Figure 2, the exposure is not uniform with peak exposures in the regions around RX J1713.73946 and the Galactic center. This irregular exposure results in an uneven sensitivity for the detection of new sources. The sensitivity of the survey in Galactic latitude (averaged over the whole longitude range) is shown in Figure 3. For 1N5 < b < 1N5 the sensitivity is rather flat at a level of 2% of the flux from the Crab Nebula and deteriorates rapidly at >2 from the Galactic plane. The flux sensitivity above 200 GeV is 50% worse for a source with photon index ¼ 3:0 compared to one with ¼ 2:0. 4. DATA ANALYSIS Following calibration of the data (Aharonian et al. 2004b), the standard H.E.S.S. event reconstruction scheme was applied to the data set. After tail-cut image cleaning and a Hillas parameterization algorithm, the direction of the shower axis was reconstructed by intersection of the major axes of the Hillas ellipses. H.E.S.S. standard cuts (Aharonian et al. 2005a) on the width and length of these ellipses (scaled to the width and length expected from Monte Carlo simulations) were used to suppress the hadronic background. An additional cut on the image size >200 photoelectrons ( p.e.) yields optimum sensitivity for weak sources provided that source spectra are not significantly steeper than that of the Crab Nebula. They result in an angular resolution of 0N08 (68% containment radius) and an energy threshold of 250 GeV. To reduce systematic effects that affect images close to the edge of the camera, only events reconstructed within a maximum distance of 2 from the pointing direction of the system were used for this analysis. The survey region was divided (a priori) into a grid of trial source positions with grid points spaced by 0N04, a value well below the angular resolution of the instrument. For each grid point in the scan region, the on-source signal was calculated by summing up all events within a circle of radius ¼ 0N22 (or 2 ¼ 0:05 deg2), where denotes the angular distance between the reconstructed event direction and the grid point. This cut is appropriate for slightly extended sources with respect to the angular resolution of the instrument and was chosen a priori, since it is optimal for the detection of gamma-ray sources with the average observed size of Galactic SNRs (Green 2004). A search for very extended sources (relative to the angular resolution) using a larger 2 cut yields similar results. A similar search for pointlike sources with a 2 ¼ 0:01 deg2 cut revealed a subset of the sources presented here and, additionally, the pointlike source HESS J1826148, which is associated with the microquasar LS 5039. The discovery of this source is reported elsewhere (Aharonian et al. 2005g). The background for each grid point was estimated using two different techniques. The standard H.E.S.S. background estimation method counts background events in a ring around each grid point
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
781
TABLE 1 On-Source Counts, Scaled Background Level (Noff According to the Notation of Li & Ma 1983), and Gamma-Ray Excess Events for the New Sources of VHE Gamma Rays Detected in the Survey of the Galactic Plane Name
Fig. 4.—Background level estimated for each grid point, calculated using a ring background technique, plotted against the background level estimated from a template method ( Rowell 2003). The black dashed line shows the expected one-to-one correlation.
with a mean radius of 0N5 and an area 7 times that of the onregion to estimate the background level. The standard ring radius of 0N5 was increased to 0N6 (again a priori) to improve sensitivity to extended sources. Using appropriate weight factors, this background determination takes into account the difference in system acceptance and exposure for the on- and the off-regions (Aharonian et al. 2005a). An alternative technique uses the template approach, where the background level is estimated using events that are normally rejected by the cosmic-ray background rejection criterion based on the scaled image width and length, but with reconstructed directions falling within the on-source region (Rowell 2003). The normalization in this case is determined by the overall ratio of events in both cut regimes, taking into account the ratio of relative acceptances in the two regimes. Figure 4 shows the correlation of the background level estimated for each grid point from the ring background and from the template background method. The slope of the major axis of the distribution is 1.007. This shows that the background estimates are consistent within 1%. The template background seems to slightly underestimate the background in comparison to the ring background. The ring background technique should therefore give a more conservative estimation of the significances of the new sources. In the following the ring background estimation method will be used as the standard method, since it is a robust method in which linear gradients in the system acceptance cancel out. From the on-counts, the background level, and the relative background acceptance in the system, a statistical significance can be determined for each grid point in the map, testing the existence of a gamma-ray source at the position of that grid point (using the approach of Li & Ma 1983). A number of significant sources are detected, and Table 1 gives the on-counts, the background level as derived from the ring background determination method (after accounting for the difference in acceptance and in the area between the on-source and off-source regions), and the resulting number of gamma-ray excess events. Since the map contains a large number of test positions, this significance has to be corrected according to a ‘‘trial factor.’’ This trial factor accounts for the increased probability of finding a fake signal with an increased number of test positions for which a significance is calculated. Using a Monte Carlo simulation of the technique applied to the survey data (including a realistic exposure map and randomly
HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS
J1614518...... J1616508...... J1632478...... J1634472...... J1640465...... J1702420...... J1708410...... J1713381...... J1745303...... J1804216...... J1813178...... J1825137...... J1834087...... J1837069......
On-Counts
Scaled Background
Excess
517 766 232 275 611 311 955 917 1138 887 717 406 478 622
283 356 139 175 298 210 728 728 927 508 374 232 287 328
234 410 93 100 313 101 227 189 211 379 343 174 191 294
Notes.—The on-counts are calculated by counting events within a circle of radius ¼ 0N22 centered on the position of the source. The background counts are derived by counting events in a ring around the source position with average radius 0N6 multiplied with the normalization factor 17 that takes into account the difference in the acceptance and the area between the on-source and the background region.
generating the events for each run from the acceptance function of that run), it has been found that this trial factor is somewhat lower than the number of points for which a significance is calculated, i.e., the number of grid points in the map (n ¼ 220;000), since the excess counts for adjacent grid points are correlated, given that the integration radius is larger than the grid spacing. For uncorrelated grid points one would have exactly the factor n. For the calculation of post-trial significances we conservatively assume that the number of trials is equal to the number of bins n. Figure 5 shows the distribution of significances for all grid points in the map (solid black line). The large positive tail to the
Fig. 5.—Distribution of significances in the survey region (solid black line). For comparison, the distribution of significances in a region without gammaray sources is shown as a dashed blue line. The distribution of significances for a Monte Carlo simulation of the survey without gamma-ray sources is shown as red filled circles and follows a Gaussian distribution with zero mean and unit width. A Gaussian fit to the dashed blue distribution yields a mean of 0:050 0:003 with an rms of 1:027 0:001, indicating a small systematic error in the determination of the significances at the level of a few percent. The dashed blue line and the red data points have been scaled to the peak height of the black distribution.
782
AHARONIAN ET AL.
Vol. 636
Fig. 6.—Significance map of the H.E.S.S. Galactic plane survey in 2004, including data from reobservations of source candidates detected in the original scan and observations of the known gamma-ray emitter RX J1713.73946 and the Galactic center region. The typical energy threshold for this map is 250 GeV. The on-source counts for each grid point are integrated in a circle of radius ¼ 0N22. The background for each grid point has been derived using a ring of mean radius 0N6 and an area 7 times that of the on-source circle. The labels indicate the gamma-ray sources described in this work, along with the known gamma-ray sources RX J1713.73946 ( HESS J1713397), G0.9+0.1 ( HESS J1747281), and the Galactic center ( HESS J1745290). The numbers in the map give the post-trial significances of the gamma-ray sources. The significance scale is truncated at 18 ; the signals from the Galactic center and RX J1713.73946 exceed this level.
distribution can be attributed to the existence of gamma-ray sources. The much smaller tail of negative significances is due to the existence of gamma-ray sources in the regions used for background estimation. This effect has been minimized in a second iteration in which regions of significant gamma-ray excess are excluded as background regions. Also shown in dashed blue is the distribution of significances for a region in the map without significant sources (l ¼ 0 20 , 0N5 < b < 1N5). This distribution has been scaled to have the same peak height as the distribution of significances for all grid points (black). A Gaussian fit to this distribution yields a mean of 0:050 0:003 with an rms of 1:027 0:003, in tolerable agreement with the expected normal Gaussian distribution given the possibility of undetected gammaray emission in the on-source or off-source regions. The distribution of significances for a Monte Carlo simulation of the survey without gamma-ray sources, using the same analysis procedure as for data, is shown with red filled circles. This distribution has also been scaled to have the same peak height as the distribution
of significances for all grid points (black) and is consistent with a normal Gaussian, demonstrating that the analysis technique behaves as expected. A post-trial significance S2 (the effective detection significance) was calculated from the pre-trial significance S1 by converting S1 into a probability, recalculating the probability of the occurrence of one or more such events given the number of trials, and converting this probability back into a significance. Grid points for which the post-trial significance is above 4.0 (corresponding to 6.3 before trials) are considered as likely sources of VHE gamma rays and are included in the following discussion. The probability of a single fake signal at this level is 3 ; 105 in the whole survey. Figure 6 shows the map of significances before accounting for the trial factor with the background estimated from the ring technique. The strong gamma-ray sources RX J1713.7 3946 and HESS J1745290 clearly stand out, partly because of the high exposure in these regions. Also visible at a much fainter level is the recently published new gamma-ray source G0.9+0.1
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
783
TABLE 2 Significances for the New Sources of VHE Gamma Rays Detected in the Survey of the Galactic Plane Best-Fit Position Significance a
Name HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS
J1614518y........... J1616508y........... J1632478............ J1634472............ J1640465y........... J1702420............ J1708410............ J1713381............ J1745303............ J1804216y........... J1813178y........... J1825137y........... J1834087y........... J1837069y...........
l (deg)
b (deg)
R.A. (deg)
Decl. (deg)
S1
S2
S3
Live Time (hr)
331.52 332.39 336.38 337.11 338.32 344.26 345.67 348.65 358.71 8.40 12.81 17.82 23.24 25.18
0.58 0.14 0.19 0.22 0.02 0.22 0.44 0.38 0.64 0.03 0.03 0.74 0.32 0.11
243.58 244.10 248.04 248.74 250.18 255.69 257.06 258.49 266.26 271.13 273.40 276.51 278.69 279.41
51.82 50.90 47.82 47.27 46.53 42.07 41.08 38.20 30.37 21.70 17.84 13.76 8.76 6.95
11.2 17.1 6.6 6.4 14.5 6.3 8.4 6.3 6.3 13.9 14.4 9.5 9.5 13.2
10.1 16.3 4.5 4.2 13.6 4.0 6.8 4.0 4.0 13.0 13.5 8.1 8.1 12.2
7.5 9.7 ... 1.8 14.5 4.1 6.9 1.4 1.8 7.7 15.3 3.7 5.7 9.9
9.8 10.2 4.5 6.6 14.3 5.7 37.4 37.3 35.3 15.7 9.7 8.4 7.3 7.6
Notes.—Source positions are derived from a fit to the gamma-ray counts in fine angular bins and do not rely on the coarser spaced search grid. S1 stands for the pre-trial significance (using a 2 cut of 0.05 deg2) for the peak positions in the map. S2 shows the post-trial significance for this position. S3 shows the post-trial significance for this position for a point-source cut (point ¼ 0N1). a The sources that are marked with a dagger have been previously reported in Aharonian et al. (2005f ).
(Aharonian et al. 2005b), which was serendipitously detected in the H.E.S.S. observations of the Galactic center region. In addition to these three known objects, 14 new sources of VHE gamma rays are detected at a significance level above 4 (post-trials), the 8 most significant of which have been discussed in a previous publication (Aharonian et al. 2005f ). Table 2 shows the pre-trial significance for the peak positions in the map for the new sources in the sixth column (S1) and the significance after accounting for the number of trials in the seventh column (S2). The eighth column (S3) shows the post-trial significance applying a cut appropriate for point sources (point ¼ 0N1) instead of the extended source cut ¼ 0N22. Only two of the new sources (HESS J1640465 and HESS J1813178) show a nonmarginal increase in significance using the point-source cut. The extension of the new sources is discussed in the next section. It should be noted that the five sources with detection significances S2 below 5 should be considered as source candidates, rather than firmly established sources. Reobservation of these sources will be necessary to confirm these marginal detections. 5. CHARACTERISTICS OF THE NEW SOURCES This section describes the main results and the analysis techniques that were applied to provide information about position, extension, and spectra of the new VHE gamma-ray sources. 5.1. Position and Morphology A detailed spectral and positional analysis has been performed on the 14 sources listed in Table 2. For this purpose, the distribution of gamma-ray candidates was determined for fine angular bins of width 0N005, to allow for a more accurate determination of the source position. For the position and approximate size determination of each source, the spatial distribution of events in the (uncorrelated) bins was fitted to a model of a twodimensional Gaussian gamma-ray brightness profile of the form / exp ( 2 /22source ), convolved with the PSF of the instrument, and an added background parameterization. To properly handle the small number of events in the bins, a likelihood fit appropriate for Poisson statistics was applied in this procedure. To
test for possible elongation of each emission region, an elongated 0 in each dimension Gaussian with independent source and source and a free orientation angle ! (measured counterclockwise from the positive Galactic latitude axis) was also fitted. The source is flagged as elongated if the error range in best-fit minor-axis 0 is outside that of the best-fit major-axis source. This fitting source procedure has been tested using Monte Carlo simulations under realistic conditions. The reanalysis of the source positions presented here in comparison to Aharonian et al. (2005f ) yields a maximum shift of 20 –30 for the most extended of these sources, reflecting the more detailed study presented here. The new positions are consistent within errors with the previously published positions. Also the previously reported extensions are confirmed by the more detailed analysis presented here, with the exception of HESS J1834087, where the extension was overestimated due to the non-Gaussian shape of the source and is now determined to be 0N09 rather than 0N20. The system PSF (see Fig. 1) has been derived from Monte Carlo simulations as a function of the zenith angle of observations and of the offset of a source in the field of view. This model PSF is consistent with the excesses observed for the pointlike Crab Nebula and the active galaxy PKS 2155304 (Aharonian et al. 2005a). Table 3 shows the best-fit position, the best-fit Gaussian equivalent 0 , and the orientation ! (in case of elongasizes source and source tion) of the 14 sources. To quantify the quality of the fit, the last column of Table 3 shows the normalized 2 between the model source shape and the measured radial profile for each source. Figure 7 shows the distribution of sizes for the new sources. For the elongated sources, the arithmetic mean of source and 0 has been filled into the histogram. All but two of the source sources appear significantly extended above a level of three standard deviations. Four out of 14 appear significantly elongated, whereas the remaining 10 new sources are compatible with a radially symmetric shape. The radial excess distributions of all sources are given individually in x 5.3. 5.2. Spectra Using the best-fit position, a spectral analysis was performed, again assuming a Gaussian emission region and using a cut of
784
AHARONIAN ET AL.
Vol. 636
TABLE 3 Position and Sizes of the New Sources of VHE Gamma Rays Detected in the Survey of the Galactic Plane Best-Fit Position
Name J1614518......... J1616508......... J1632478......... J1634472......... J1640465......... J1702420......... J1708410......... J1713381......... J1745303......... J1804216......... J1813178......... J1825137......... J1834087......... J1837069.........
l (deg) 331.52 332.391 336.38 337.11 338.316 344.26 345.672 348.65 358.71 8.401 12.813 17.82 23.24 25.185
b (deg) 0.03 0.014 0.04 0.05 0.007 0.04 0.013 0.03 0.04 0.016 0.005 0.03 0.02 0.012
0.58 0.138 0.19 0.22 0.021 0.22 0.438 0.38 0.64 0.033 0.0342 0.74 0.32 0.106
0.02 0.013 0.03 0.04 0.007 0.03 0.010 0.03 0.05 0.018 0.005 0.03 0.02 0.016
source (deg)
0 source (deg)
0.15 0.02 ... 0.06 0.04 ... ... ... ... ... 0.09 0.04 ... ... ... ... 0.05 0.02
0.23 0.136 0.21 0.11 0.045 0.08 0.054 0.06 0.21 0.200 0.036 0.16 0.09 0.12
0.02 0.008 0.05 0.03 0.009 0.04 0.012 0.04 0.06 0.010 0.006 0.02 0.02 0.02
! (deg) 49 ... 21 ... ... ... ... ... 54 ... ... ... ... 149
2norm
10
1.44 0.82 1.05 0.72 0.91 1.38 0.86 1.04 1.17 1.34 0.94 1.03 1.16 1.52
13
7
10
0 Notes.—Galactic longitude l and latitude b, the rms source size source and source , and the orientation angle ! counterclockwise with respect to the positive Galactic latitude axis are given in degrees. Parameter 2norm denotes the normalized 2, with 49 degrees of freedom.
2 < 2point (2source þ2PSF )/2PSF , approximating the H.E.S.S. PSF as a two-dimensional Gaussian with PSF ¼ 0N08 and point ¼ 0N1 being the optimal cut for point sources. Such a cut provides optimum significance for a weak Gaussian-shaped source on a uniform background and hence the best possible spectral determination. For the elongated sources, the larger of the two values source and 0 was used. The energy-dependent efficiency of this cut is source estimated by convolving the H.E.S.S. PSF with a Gaussian of size source. This introduces an additional systematic error in the absolute flux determination since it implicitly assumes that the emission region of the source follows a Gaussian distribution and that the morphology of the source does not show any energy dependency. The typical overall systematic error in the determination of the integral flux level is estimated to be 30%. The standard H.E.S.S. spectral analysis technique is described in Aharonian et al. (2005a). To avoid systematic errors in the spectral determination arising from different camera acceptances across the H.E.S.S. field of view, the spectral background determination employs regions of the same size and shape as the on-region, but displaced on a
ring in the field of view around the camera center. The ring was chosen such that the off-regions have the same angular distance to the center as the on-region. This approach ensures that background events are taken at the same zenith and offset angles. To keep as many low-energy gamma rays as possible, a looser size cut of 80 photoelectrons is used in the spectral analysis. This cut allows us to reach a lower energy threshold (150 GeV ) at the expense of reduced sensitivity for hard-spectrum sources and a 30% poorer angular resolution. Flux points below a significance level of 2 standard deviations were discarded in the spectral analysis, and the binning was doubled at high energies for weak sources. Table 4 lists the resulting photon index and flux above 200 GeV for a power-law fit dN /dE ¼ F0 E . The typical systematic error on these photon indices is estimated to be 0.2. It should be noted that the fluxes reported here are in general different TABLE 4 Results of a Power-Law Fit dN /dE ¼ F0 E for the 14 Sources of VHE Gamma Rays
Name HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS HESS
Fig. 7.—Distribution of the rms angular size source of the 14 VHE gammaray sources. For the elongated sources, the arithmetic mean of source and 0 is used. The mean of this distribution is 0N1 with an rms of 0N05. source
J1614518..... J1616508..... J1632478..... J1634472..... J1640465..... J1702420..... J1708410..... J1713381..... J1745303..... J1804216..... J1813178..... J1825137..... J1834087..... J1837069.....
2.46 2.35 2.12 2.38 2.42 2.31 2.34 2.27 1.82 2.72 2.09 2.46 2.45 2.27
0.20 0.06 0.20 0.27 0.15 0.15 0.11 0.48 0.29 0.06 0.08 0.08 0.16 0.06
Flux > 200 GeV (1012 cm2 s1)
Cut (deg)
2/dof
0.40 0.26 0.36 0.23 0.16 0.20 0.17 0.17 0.36 0.36 0.15 0.30 0.20 0.23
0.6/3 5.5/5 7.9/2 5.8/5 3.7/4 7.8/6 6.3/3 1.8/2 4.4/3 6.9/11 5.5/11 6.2/8 3.5/4 12.6/10
57.8 43.3 28.7 13.4 20.9 15.9 8.8 4.2 11.2 53.2 14.2 39.4 18.7 30.4
7.7 2.0 5.3 2.6 2.2 1.8 0.7 1.5 4.0 2.0 1.1 2.2 2.0 1.6
Notes.—The cut used for the spectral analysis is given in degrees. The errors given on the photon index and flux are statistical only. The typical systematic errors are 0.2 in the photon index and 30% in the absolute flux. To derive a combined error on the photon index, the systematic and statistical errors should be added in quadrature.
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
Fig. 8.—Distributions of the photon index of the new sources. The mean photon index is 2.32 with an rms of 0.2.
from the fluxes reported in Aharonian et al. (2005f ) (especially for the larger sources) since in the present work the flux is determined from a larger region, adapted to encompass a larger fraction of the emission region, whereas in Aharonian et al. (2005f ), a fixed small 2 ¼ 0:02 deg2 cut was used for all sources. Figure 8 shows the distribution of photon indices for the 14 new sources. The distribution exhibits a mean of 2.32, marginally consistent with predictions by models of shock wave acceleration in SNRs (see, e.g., Berezhko & Vo¨lk 1997, 2000) and as expected for the source spectrum of Galactic cosmic rays (see, e.g., Jones et al. 2001). The distribution has an rms of 0.2, comparable with the statistical errors on the measurements. 5.3. Individual Gamma-Ray Sources This section describes the 14 VHE gamma-ray sources detected in the Galactic plane survey individually along with a brief description of possible counterparts at other wavelengths. Excess maps are shown that have been smoothed with a two-dimensional Gaussian to reduce the effect of statistical fluctuations. The resulting count map is in units of integrated excess counts within the smoothing radius. The rms radius of the Gaussian used for the smoothing is adapted for each source according to the avail-
785
able photon statistics. This radius is given in the figure captions and is shown as a dotted black line in the inset in the lower right part of each excess map along with the PSF of the instrument, smoothed in the same way as the excess map. Note that the maps have different scales; therefore, the white line in the upper right corner indicates a 0N2 scale. Further, it should be noted that the color scale covers different intervals for different maps. Also shown are possible counterparts to the VHE gammaray emission: SNR from Green (2004) (with white circles illustrating their nominal radii), pulsars from the ATNF pulsar catalog ( Manchester et al. 2005; white triangles), hard X-ray sources detected by the International Gamma-Ray Astrophysics Laboratory (INTEGRAL) satellite (white stars), and unidentified EGRET sources from the third EGRET catalog ( Hartman et al. 1999) (showing the 95% error contour for the position of the source in dashed white). In some maps contours of radio and/or X-ray emission are also shown and discussed individually. The 20 cm Very Large Array (VLA) radio contours are taken from White et al. (2005). The X-ray ASCA data show smoothed count maps obtained with the GIS instrument in the 0.7–10 keV band. The best-fit source position of the H.E.S.S. source is shown with error bars in black, and the best-fit (Gaussian) source size is shown as a black dashed circle or ellipse. The radial distribution of gamma-ray excess with respect to the best-fit source position is shown for each source binned appropriately for the available photon statistics; in each plot the number of reconstructed events in bins of the squared angular distance (2) is shown along with the best-fit radially symmetric model (solid black line) and the PSF of the instrument (dashed red line), derived from Monte Carlo simulations. For all sources the reconstructed differential energy spectra are shown together with the best-fit power law (solid line). 5.3.1. HESS J1614518
This source (see Fig. 9) is located in a region relatively devoid of counterpart candidates. The source exhibits elliptical emission with a semimajor axis of 14 0 1 0 and a semiminor axis of 9 0 1 0 . HESS J1614518 is the brightest of the new sources, detected with a flux corresponding to 25% of that from the Crab Nebula above 200 GeV. The spectrum is well fitted by a power
Fig. 9.—Left: Smoothed excess map (smoothing radius 0N09) of the region of HESS J1614518. Right: Energy spectrum (top) and 2 plot (bottom).
786
AHARONIAN ET AL.
Vol. 636
Fig. 10.—Left: Smoothed excess map (smoothing radius 0N06) of the region surrounding HESS J1616508 (left-hand source), along with nearby pulsars and SNRs that were considered as counterparts. HESS J1614518 is also visible in this map (right-hand source). Right: Energy spectrum (top) and the radial distribution (2) plot (bottom).
law of photon index 2:46 0:20 (see Fig. 9, top right). A nearby Chandra observation (Kastner et al. 2003) covered an area more than 100 away from the best-fit position of HESS J1614518 and showed no evidence for X-ray emission from the direction of the H.E.S.S. source. The poor quality of the fit to the radial distribution of HESS J1614518 as is shown in Figure 9 (bottom right) indicates that the source is not well described by a symmetrical Gaussian profile. 5.3.2. HESS J1616508
HESS J1616508 (see Fig. 10), with a flux of 19% of the Crab Nebula above 200 GeV, is also one of the brighter new sources of VHE gamma rays. Its spectrum can be fitted by a power law of photon index 2:35 0:06 as shown in Figure 10 (top right). The emission region has an rms size of 160 and does not have a good positional match to any plausible counterpart at other wavelengths. It is located in a region close to the known SNRs G332.40.4 ( RCW 103) and G332.4+0.1 (Kes 32). The region has been intensively studied with X-ray satellites like Chandra ( Vink 2004) and XMM-Newton, with the position of HESS J1616508 mostly outside or at the edge of the field of view where the sensitivity is strongly decreased. No extended X-ray source that lines up with the VHE gamma-ray emission has been found. Another interesting object in this region is the energetic pulsar PSR J16175055 (Torii et al. 1998), a young X-ray–emitting pulsar that was proposed to emit TeV gamma rays (Hirotani 2001). Approximately 1.2% of its spin-down luminosity would suffice to power the observed (0.2–10 TeV ) VHE gamma-ray flux, assuming that the H.E.S.S. source is located at the pulsar distance. Since HESS J1616508 does not line up with this object, an asymmetric (yet undetected) PWN powered by the pulsar would be required to explain the TeV emission. 5.3.3. HESS J1632478
The centroid positions of this object (right-hand source in Fig. 11, top) and of the close-by HESS J1634472 are spatially separated by less than 450 . HESS J1632478 has an elongated
shape with a semimajor axis of 12 0 3 0 and a semiminor axis of 3A6 2A4. Its flux in VHE gamma rays above 200 GeV corresponds to 12% of the flux from the Crab Nebula, and the spectrum can be fitted by a power law of photon index 2:12 0:20 (see Fig. 11, bottom right). A positional coincidence of HESS J1632478 exists with the hard X-ray source IGR J16320 4751 discovered by the INTEGRAL satellite (Tomsick et al. 2003) in the energy range above 15 keV. In softer X-rays between 2 and 10 keV, this object was also strongly detected in an XMM-Newton observation ( Rodriguez et al. 2003) and found to be coincident with the ASCA source AX J1631.94752, detected in the ASCA Galactic plane survey (Sugizaki et al. 2001). This source is believed to belong to a new class of heavily absorbed high-mass X-ray binary systems, with an equivalent absorption column density of (9:2 1:1) ; 1022 cm2 in this case. The extension of HESS J1632478 does not plead in favor of an association with IGR J163204751. One should note that the probability of spurious associations is increased in this region of the Galactic plane (the 3 kpc arm tangent region) due to the high density of absorbed INTEGRAL sources. Another potential counterpart is the ASCA source AX J1632524746. 5.3.4. HESS J1634472
The size of the HESS J1634472 (left-hand source in Fig. 11, top) emission region is 6A6 1A8, and its flux above 200 GeV is close to 6% of the Crab flux with a photon index for the energy spectrum of 2:38 0:26 (see Fig. 11, middle left). HESS J1634 472 has no direct positional counterpart, although it is interesting to note that yet another INTEGRAL source IGR J163584726 is located close by (Revnivtsev et al. 2003). This source presumably belongs to the class of highly absorbed X-ray binary systems, like IGR J163204751 (discussed in the previous paragraph). The source was detected in the 2–10 keV X-ray band with the Chandra satellite (Patel et al. 2004). The column density determined for this object is NH ¼ 3:3 ; 1023 cm2, and the luminosity of this object in the 2–10 keV band is 1036 ergs s1. Another potential counterpart of HESS J1634472 is the recently discovered SNR
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
787
Fig. 11.—Top: Smoothed excess map (smoothing radius 0N12) of the region of the two close-by H.E.S.S. sources HESS J1634472 (left-hand source) and HESS J1632478 (right-hand source), along with nearby pulsars, SNRs, and INTEGRAL sources that were considered as counterparts. It should be noted that the tail of HESS J1632478 to negative Galactic latitude is statistically not significant. The panels in the middle row show the spectra (left: HESS J1634472; right: HESS J1632478), while the panels in the bottom row show the corresponding 2 plots.
candidate G337.2+0.1 (Combi et al. 2005), coincident with the ASCA source AX J1635.94719. However, the distances of the two objects to HESS J1634472 render a correlation unlikely. 5.3.5. HESS J1640465
As can be seen from the radial distribution of this source (see Fig. 12, bottom right), HESS J1640465 is marginally extended with respect to the PSF of the instrument. It is one of the few examples of an excellent spatial correlation between the VHE gamma-ray emission and an SNR (G338.30.0), known from ra-
dio observations (Green 2004). G338.30.0 was observed in the Molonglo Galactic plane survey at 843 MHz (shown in green contours in Fig. 12, left) and has been reported to be a broken shell SNR lying on the edge of a bright H ii region ( Whiteoak & Green 1996). The region bridging this SNR and the close-by SNR G338.5+0.1 is known to contain dense H ii regions. The relation of surface brightness to distance yields a distance of 8.6 kpc, which would place the SNR in the 3 kpc arm tangent region. The ASCA source AX J1640424632 with a photon index of 2:98þ1:13 0:89 and a flux level of 1:21 ; 1012 ergs cm2 s1, detected during
788
AHARONIAN ET AL.
Vol. 636
Fig. 12.—Left: Smoothed excess map (smoothing radius 0N05) of the region surrounding HESS J1640465 along with nearby SNRs that were considered as counterparts as discussed in the text. The green contours indicate the radio emission detected from the region surrounding G338.30.0 in the Molonglo Galactic plane survey at 843 MHz (Green et al. 1999). The dashed white contours show the 95% positional error of the unidentified EGRET source 3EG J16394702. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
the ASCA Galactic plane survey (Sugizaki et al. 2001), has been identified with G338.30.0. It should be noted that the position of HESS J1640465 is compatible with the position of the unidentified EGRET source 3EG J16394702 (Hartman et al. 1999). The 95% positional confidence contour of 3EG J1639 4702 is shown in Figure 12 (left). The distance of HESS J1640 465 to this EGRET source is 340 . The differential energy spectrum of HESS J1640465 can be well fitted by a power law with photon index 2:42 0:14 at a flux close to 9% of that from the Crab Nebula above 200 GeV (see Fig. 12, top right). A simple power-law extrapolation of the VHE gamma-ray spectrum to 1 GeV matches the EGRET flux above that energy.
5.3.6. HESS J1702420
For HESS J1702420 ( Fig. 13) no significant evidence for extension of the emission region can be derived beyond a level of 2 standard deviations for the measured source size of 4A8 2A4. The flux above 200 GeV of this source corresponds to 7% of the Crab flux, and the energy spectrum can be fitted by a power law of photon index 2:31 0:15 (see Fig. 13, top right). The closeby pulsar PSR J17024128 is energetic enough to account for the observed VHE gamma-ray emission by converting roughly 14% of its spin-down luminosity but would need an asymmetric PWN to be responsible for the observed VHE gamma-ray
Fig. 13.—Left: Smoothed excess map (smoothing radius 0N10) of the region surrounding HESS 1702420 along with nearby pulsars and SNRs that were considered as counterparts as discussed in the text. The tail extending to positive Galactic longitude and latitude is not statistically significant. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
789
Fig. 14.—Left: Smoothed excess map (smoothing radius 0N05) of the region surrounding HESS 1708410 along with a nearby SNR that was considered as a counterpart. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
emission. No asymmetric PWN has been detected from this pulsar, however, which renders an association unlikely. The nearby SNR G344.70.1 is too small and not positionally coincident with the VHE gamma-ray source. Therefore, we cannot identify a plausible counterpart for this object. 5.3.7. HESS J1708410
This VHE gamma-ray source (Fig. 14) of size 3A2 0A7 is located very close to SNR RX J1713.73946 and therefore its observation live time (within 2 of the pointing position of the instrument) amounts to 37.2 hr, well beyond typical exposures in the H.E.S.S. survey. The source exhibits an energy spectrum with a power-law index of 2:34 0:11, and the flux above 200 GeV is less than 4% of the flux from the Crab Nebula (see Fig. 14, top right). As can be seen from Figure 14 (left), there is no close-by plausible counterpart. A nearby XMM-Newton observation of the SNR G345.70.2 has revealed no X-ray counterpart to the VHE gamma-ray source, but it should be noted that the offset of HESS J1708410 from the center of the XMM-Newton field of view was 150 , positioning it close to the edge of the field of view. As can be seen from the radial distribution, shown in Figure 14 (bottom right), the source is only marginally extended with respect to the PSF of the instrument. 5.3.8. HESS J1713381
This source (Fig. 15) is the second of the 14 VHE sources that show no significant evidence beyond 2 standard deviations for an extension of the emission region. The energy spectrum of HESS J1713381 is consistent with a power law with photon index 2:2 0:5. Its flux of 1.8% of the flux from the Crab Nebula above 200 GeV makes this object the weakest of the new sources. Its location is very close to the strong VHE gamma-ray source SNR RX J1713.73946, and it was therefore in the field of view of most pointed observations by H.E.S.S. on this object. The correspondingly deep exposure of 37 hr at offsets of less than 2 from the center of the field of view made it possible to detect this object. As can be seen from Figure 15 (left), it coincides with part of the unusual SNR complex CTB 37. The radio contours of the CTB 37 complex as taken from the Molonglo Galactic plane
survey at 843 MHz are shown in white (Green et al. 1999). There is clear positional match between the source HESS J1713381 and the SNR G348.7+0.3 (CTB 37B). Also visible in Figure 15 (right) is a weaker marginal source consistent with the position of G348.5+0.1 (CTB 37 A). The marginal detection of emission from this source is however not significant after accounting for all trials for the analysis of extended sources in this survey as described above. Kassim et al. (1991) describe CTB 37A as two SNRs overlapping in projection: with the partial shell of G348.7+ 0.1 in the west and the SNR G348.50.0 to the east. Both SNRs in CTB 37A appear to be interacting with a system of molecular clouds with densities of 100–1000 cm3 (Reynoso & Mangum 2000). The SNR CTB 37B is less well studied but appears to share the dense environment of CTB 37A. Evidence for X-ray emission from the CTB 37 complex can also be seen in archival ASCA data. 5.3.9. HESS J1745303
This object ( Fig. 16) has a direct positional coincidence with an unidentified EGRET source (3EG J17443011; Hartman et al. 1999). Since it is within the field of view of the Galactic center, HESS J1745303 has a large observation live time of 35.3 hr. Its flux level corresponds to 5% of the flux from the Crab Nebula above 200 GeV, and the emission exhibits a hard differential flux of index 1:8 0:3. Its emission region has an elliptical shape with a semimajor axis of 13 0 4 0 and a semiminor axis of 5A4 2A4 as is shown in Figure 16 (left). The distance of HESS J1745303 to the centroid of the EGRET source 3EG J17443011 is 100 , which is well within the 95% uncertainty level of the EGRET position. A simple power-law extrapolation of the VHE gammaray spectrum to energies above 1 GeV is an order of magnitude below the EGRET flux above that energy [F(>100 MeV) ¼ (64 7) ; 108 cm2 s1]. Another potential counterpart is the pulsar PSR B174230, but essentially 100% of its spin-down power would be required to account for the entire VHE gammaray emission. 5.3.10. HESS J1804216
This source ( Fig. 17), with a size of approximately 220 , is the largest of the new emission regions of VHE gamma rays.
790
AHARONIAN ET AL.
Vol. 636
Fig. 15.—Left: Smoothed excess map (smoothing radius 0N07) of the region surrounding HESS J1713381 along with nearby SNRs that were considered as counterparts as discussed in the text. In the lower right corner of this excess map, the strong VHE emission from SNR RX J1713.73946 is apparent. The white contours show the radio emission as seen in the Molonglo plane survey at 843 MHz (Green et al. 1999). The marginal excess in positional agreement with G348.5+0.1 (CTB 37A) is not significant after accounting for all trials for the analysis presented here. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom). Also in the radial distribution, some indication for an excess 0N4 away from HESS J1713381 can be seen, due to the marginal excess at the position of CTB 37A. The dashed white contours show the 95% positional error of the unidentified EGRET source 3EG J17143857.
With a flux of nearly 25% of the flux from the Crab Nebula above 200 GeV, it is also the brightest of the new H.E.S.S. sources. The photon index of 2:72 0:06 makes it one of the softest sources (see Fig. 17, top right). As can be seen from Figure 17 (left), the emission region does not perfectly line up with any known object, although it could be associated with the southwestern part of the shell of the SNR G8.70.1 of radius 260 . From CO observations, the surrounding region is known to be associated with molecular gas where massive star formation takes place (Ojeda-May et al.
2002). The white contours in Figure 17 (left) show soft X-ray emis¨ gelman 1994), sion as detected by the ROSAT satellite (Finley & O while the black contours show the 20 cm radio emission as recorded with the VLA. Another plausible association of HESS J1804216 is the young Vela-like pulsar PSR J18032137 with a spin-down age of 16,000 yr (Kassim & Weiler 1990). The required efficiency of this pulsar’s spin-down luminosity to power the observed emission in gamma rays is only 3%, and it could therefore easily account for the new source.
Fig. 16.—Left: Smoothed excess map (smoothing radius 0N10) of the region surrounding HESS J1745303 along with nearby pulsars and SNRs that were considered as counterparts as discussed in the text. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
791
Fig. 17.—Left: Smoothed excess map (smoothing radius 0N06) of the region surrounding HESS J1804216 along with nearby pulsars and SNRs that were considered as counterparts as discussed in the text. The white contours show the X-ray emission from G8.70.1 as detected by the ROSAT satellite, while the black contours show the 20 cm radio emission detected by the VLA. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
5.3.11. HESS J1813178
With a moderate flux of approximately 6% of the Crab Nebula above 200 GeV, this source (see Fig. 18) is among the more compact of the new sources with an extension of 2A2. When using a large 2 cut of (0N4)2, the normalization of the power-law fit is 25% higher, suggesting a faint non-Gaussian tail in the radial brightness profile. The source exhibits a rather hard photon index of 2:09 0:08 (see Fig. 18, top right). In the previous publication (Aharonian et al. 2005f) this source was still considered as unidentified. As it is clear from Figure 18 (left), there is a compelling positional coincidence between this source and the X-ray source AX J1813178 (Brogan et al. 2005) (shown by white contours in Fig. 18 for the 0.7–10 keV band). This ASCA source
is one of the bright sources detected in the ASCA Galactic plane survey with a very hard energy spectrum. This object was also recently detected in the INTEGRAL Galactic plane survey in the 20–100 keV energy band (Ubertini et al. 2005). The region surrounding HESS J1813178 was also observed with the VLA, and 20 cm radio contours are shown in black in Figure 18. There is a faint radio source (G12.820.02) with a shell-like structure visible at the position of HESS J1813178 as also recently reported (Brogan et al. 2005; Helfand et al. 2005). This indicates a probable association with a shell-type SNR. Additionally HESS J1813178 lies 100 from the center of the radio source W33. W33 extends over 150 and has a compact radio core (G12.80.2) that is 10 across (Haschick & Ho 1983), located 150 away from HESS J1813178. The W33 region shows signature of an ultracompact
Fig. 18.—Left: Smoothed excess map (smoothing radius 0N04) of the region surrounding HESS J1813178 along with nearby pulsars and SNRs that were considered as counterparts as discussed in the text. The white contours show the smoothed X-ray count map for the region surrounding AX J1813178 obtained by ASCA with the GIS instrument in the 0.7–10 keV band, while the black contours show 20 cm radio emission as observed by the VLA. The strong radio source to negative Galactic longitudes from HESS J1813178 is connected to the W33 complex and extends farther than shown here. The excess seen to negative Galactic longitudes at the edge of the ASCA GIS field of view is likely due to contamination from GX 13+1. The radial (bottom right) distribution (2) plot shows that the source is marginally extended with respect to the PSF of the H.E.S.S. system.
792
AHARONIAN ET AL.
Vol. 636
Fig. 19.—Left: Smoothed excess map (smoothing radius 0N09) of the region surrounding HESS J1825137 along with the nearby pulsar that was considered as a counterpart as discussed in the text. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
H ii region (Churchwell 1990) and contains methanol, hydroxyl, and water masers and other tracers of recent star formation, which could act as target material for the generation of VHE gamma rays but is not coincident with the H.E.S.S. source. 5.3.12. HESS J1825137
This source has an energy spectrum with a photon index of 2:46 0:08 and a flux of VHE gamma rays of approximately 17% of the flux from the Crab Nebula above 200 GeV (see Fig. 19, top right). It is approximately radially symmetric, and the emission region can be best described by a Gaussian of width 100 . The position of HESS J1825137 is approximately 110 to negative Galactic longitudes of the young pulsar PSR B182313 (PSR
J18261334), which makes an association of the VHE gammaray source with the pulsar possible. The pulsar PSR B182313 is a 101 ms evolved Vela-like pulsar at a distance of 3:9 0:4 kpc, and at this distance the VHE luminosity of HESS J1825137 would only be 2% of its spin-down power. The region surrounding the pulsar has been previously observed at large zenith angles by the Whipple collaboration ( Hall et al. 2003), who observed a 3.1 excess also to negative Galactic longitudes with respect to the pulsar position. X-ray observations with the XMM-Newton satellite show a 50 diameter diffuse emission region extending also asymmetrically to negative Galactic longitudes from the pulsar (Gaensler et al. 2003). If the VHE gamma-ray source is indeed associated with this PWN, the VHE emission region would
Fig. 20.—Left: Smoothed excess map (smoothing radius 0N08) of the region surrounding HESS J1834087 along with nearby pulsars and SNRs that were considered as counterparts as discussed in the text. The positional coincidence with SNR G23.30.3 (W41) is evident; the black contours show 20 cm radio emission recorded by the VLA (White et al. 2005). Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
793
Fig. 21.—Left: Smoothed excess map (smoothing radius 0N05) of the region surrounding HESS J1837069 along with nearby SNRs that were considered as counterparts as discussed in the text. The black contours show the smoothed X-ray count map for the region surrounding AX J1838.00655 obtained by ASCA with the GIS instrument in the 0.7–10 keV band. The positional coincidence of HESS J1836069 with the ASCA source AX J1838.00655 is evident. Right: Differential energy spectrum (top) and the radial distribution (2) plot (bottom).
extend much farther to the south than in X-rays. PSR B182313 was also proposed to power the close-by unidentified EGRET source 3EG J18261302 ( Nolan et al. 2003). The distance of the centroid of this EGRET source to HESS J1825137 is 430 . As illustrated in Figure 19 (left), the new VHE gamma-ray source is well compatible with the EGRETsource within the 95% positional error of EGRET, and a simple power-law extrapolation of the H.E.S.S. energy spectrum to energies above 1 GeV gives a similar flux as derived from the EGRET spectrum above this energy. An in-depth study of HESS J1825137 and discussion about the corresponding PWN model will be given in a separate paper (Aharonian et al. 2005e). 5.3.13. HESS J1834087
HESS J1834087 ( Fig. 20) has a size of 120 and a gammaray flux of 8% of the flux from the Crab Nebula above 200 GeV. Its energy spectrum exhibits a photon index of 2:45 0:16 (see Fig. 20, top right). When using a large 2 cut of (0N4)2, the normalization of the power-law fit increases by 30%, suggesting a deviation of the morphology from the assumed radial Gaussian brightness profile. The location of HESS J1834087 makes it another source with a compelling positional agreement with an SNR (see Fig. 20, left). The positionally coincident shell-type SNR G23.30.3 (W41) with a diameter of 270 was mapped in radio with the VLA at 330 MHz ( Kassim 1992) and at 20 cm (White et al. 2005) as shown in black contours in Figure 20 (left). It is possibly connected to the old pulsar PSR J18330827 (Gaensler & Johnston 1995). This pulsar is energetic enough to power HESS J1834087, but its distance of 240 renders an association unlikely; there is also no extended PWN detected so far. 5.3.14. HESS J1837069
This source ( Fig. 21) has an elongated shape with a semimajor axis of 7 0 1 0 and a semiminor axis of 3 0 1 0 . With a flux of 13.4% of the Crab flux above 200 GeV, the energy spectrum can be well fitted by a power law of photon index 2:27 0:06 (see
Fig. 21, top right). When using a large 2 cut of (0N4)2, the normalization of the power-law fit increases by 30%, suggesting that the assumption of a Gaussian brightness profile does not hold for this source. HESS J1837069 coincides with the southern part of the diffuse hard X-ray complex G25.5+0.0. This source, which is 120 across, was detected in the ASCA Galactic plane survey ( Bamba et al. 2003). The nature of this bright X-ray source is unclear, but it may be an X-ray synchrotron emission dominated SNR such as SN 1006 or a PWN. The brightest feature in the ASCA map (AX J1838.00655), located south of G25.5+0.0, coincides with the VHE emission. This still unidentified source was also serendipitously detected by BeppoSAX and also in the hard X-ray (20–100 keV ) band in the Galactic plane survey of INTEGRAL ( Malizia et al. 2004). 6. DISCUSSION Possible multiwavelength associations of the 14 new VHE gamma-ray sources were discussed in the previous section and are summarized in Table 5. A further qualification from tentative associations to unambiguous counterpart identifications requires spatial or morphological coincidence, a viable gamma-ray emission mechanism for the object, and consistent multiwavelength behavior according to the properties of the suggested identification. Most of the possible associations given above can be considered as plausible gamma-ray emitters, but only in a minority of these cases does a satisfactory positional coincidence exist. Four of the new sources apparently line up with SNRs. These SNRs have been detected and studied mostly in the radio domain, but in the cases of HESS J1640465 and HESS J1813178 X-ray emission was detected by ASCA, and there is marginal evidence for X-ray emission in ASCA archival data in the case of HESS J1713381. Only in the case of HESS J1834087 is there currently no evidence for associated X-ray emission. For three of the new sources, namely, HESS J1616509, HESS J1804216, and HESS J1825137, a sufficiently energetic nearby pulsar could account for the VHE gamma-ray emission,
794
AHARONIAN ET AL.
Vol. 636
TABLE 5 Summary of Possible Counterparts to the H.E.S.S. VHE Gamma-Ray Sources in the Inner Galaxy
Name
Possible Counterpart
Class
Offset (arcmin)
Distance (kpc)
Luminosity (1034 ergs s1)
J1614518.............. J1616508.............. J1632478.............. J1634472.............. J1640465.............. J1702420.............. J1708410.............. J1713381.............. J1713397.............. J1745290.............. J1745303.............. J1747281.............. J1804216.............. J1813178.............. J1825137.............. J1834087.............. J1837069..............
... PSR J16175055 IGR J163204751 IGR J163584726/G337.2+0.1 G338.30.0/3EG J16394702 ... ... G348.7+0.3 RX J1713.73946 (G347.30.5) Sgr A East /Sgr A 3EG J17443011 G0.9+0.1 G8.70.1/PSR J18032137 G12.820.02 PSR J18261334/3EG J18261302 G23.30.3 AX J1838.00655
... PWN XRB XRB/SNR SNR / UID ... ... SNR SNR SNR / BH UID PWN SNR / PWN SNR PWN/ UID SNR UID
... 10.4 3 13/10 0/34 ... ... 0 0 0 10 0 21/10.8 0 11/43 0 6
... 6.5 ... . . ./14 8.6 ... ... 10.2 1 8.5 ... 8.5 6/3.9 4 3.9/. . . 4.8 ...
... 20.2 ... . . ./28.3 16.2 ... ... 5.2 1.2 12.8 ... 4.4 16.5/7.0 3.1 6.1/. . . 4.4 ...
Notes.—The offset denotes the angular distance of the VHE gamma-ray source from the possible counterpart. Distances are taken from Green (2004), Manchester et al. (2005), and references therein. The last column gives the implied luminosity between 0.2 and 10 TeV. The three sources that were known before the H.E.S.S. survey in the inner Galaxy ( HESS J1747218, HESS J1745290, and HESS J1713 397) have been added for completeness to this table. The possible source classes have been abbreviated according to the following scheme: supernova remnant (SNR), unidentified sources ( UID), pulsar wind nebula ( PWN ), black hole candidate ( BH), X-ray binaries ( XRB).
but the offset of the sources from the pulsar positions makes these associations less firm. Three of the new VHE gamma-ray sources are consistent with the 95% positional error of unidentified EGRET sources ( HESS J1640465, HESS J1745303, and HESS J1825137). Two of the new sources are located very close to a new class of highly absorbed hard X-ray sources detected recently by INTEGRAL (HESS J1632478 and HESS J1634472). A third source (HESS J1837067) may be related to X-ray emission of unknown origin detected by ASCA and INTEGRAL, which may belong to the same class of objects. The three remaining sources (HESS J1614518, HESS J1702420, and HESS J1708410) have no plausible SNR, pulsar, or EGRET counterpart and may belong to a new source class. Significant fluxes of VHE gamma rays without accompanying X-ray and radio emission would suggest the absence of relativistic electrons and the presence of energetic nucleons. That such a population exists is already being suggested by the detection of apparently ‘‘dark’’ accelerators by the HEGRA (Aharonian et al. 2002a) and H.E.S.S. (Aharonian et al. 2005d) collaborations. However, the possibility remains that these new sources are associated with SNRs or PWNs as yet undiscovered at other wavelengths. Further multiwavelength investigations are needed to clarify this picture. The nearest energy band relative to the H.E.S.S. energy range was covered by observations with EGRET aboard the Compton Gamma Ray Observatory. Its energy coverage was primarily between 30 MeV and 10 GeV, and gamma-ray source detections above 100 MeV resulted in the third EGRET (3EG) catalog ( Hartman et al. 1999). The only Galactic sources detected above 10 GeV are pulsars, for which the sparse photons at these energies correlate with the location and the pulse profile at lower EGRET energies (Thompson et al. 2005). The bulk of photons above 10 GeV have to be attributed to Galactic (and extragalactic) diffuse emission. When comparing the EGRET sources with the H.E.S.S. Galactic plane survey result, it should be noted that even if an EGRET source spectrum is measured up to 10 GeV, more than a decade in energy is left uncovered below the H.E.S.S. energy threshold. In-
deed, only a minor fraction of the H.E.S.S. sources coincide within the considerably larger location uncertainty contours of EGRET sources. These sources are HESS J1640465, HESS J1745 303, and HESS J1825137, which are consistent with the 95% location uncertainty of the unidentified EGRET sources 3EG J16394702, 3EG J17443011, and 3EG J18261302. In total 17 EGRETsources are cataloged within the H.E.S.S. survey region. Taking into account the 95% positional error contours, they cover an area of 9.76 deg2, resulting in a chance positional coincidence of a given H.E.S.S. source with an EGRET source of 32%. A correlated Galactic latitude distribution of the EGRET and H.E.S.S. sources, concentrated near the Galactic plane, could increase this probability. Physical arguments suggest that the EGRET source population and the set of new H.E.S.S. sources might be of different kinds. While the brightest gamma-ray pulsars can be directly traced up to 20 GeV, EGRET data indicate the existence of GeV cutoffs already in the 2–10 GeV range ( Reimer & Bertsch 2001). The brightest pulsars and several unidentified EGRET sources with pulsar-like characteristics deviate from single power-law spectral representations (Bertsch et al. 2000) and are significantly better fitted when introducing high-energy spectral cutoffs. GeV cutoffs are indeed predicted from multifrequency modeling of the gammaray emission of pulsars (Cheng 2004; Baring 2004). Therefore, the emission of these unidentified EGRET sources may not extend up to the H.E.S.S. energy range. From Crab measurements by CELESTE (de Naurois et al. 2002), consecutively continuing the EGRET GeV data, a characteristic change in the emission properties of the pulsar became apparent: pulsed emission is diminished or entirely absent, leaving only a steady emission component, in the case of Crab attributed to the PWN. If the majority of EGRET sources in the Galactic plane may be identifiable with gammaray pulsars, this population will overlap with the H.E.S.S. energies only if plerionic emission components exist. For PWNs there are theoretical grounds to expect in some cases a displacement of the gamma-ray emission from the pulsar powering the nebula (Aharonian et al. 1997a; Blondin et al. 2001).
No. 2, 2006
H.E.S.S. SURVEY OF INNER GALAXY
795
Fig. 22.—Distribution of Galactic latitude of the 17 VHE gamma-ray sources as given in Table 5 (the 14 new plus the 3 previously known). The mean of the distribution is 0.17 with an rms of 0.31. On the left, the distribution is shown along with the distribution in Galactic latitude of possible counterpart populations to the VHE gamma-ray sources. SNRs from Green (2004) are shown by a dot-dashed purple line, while high spin-down luminosity pulsars (E˙ > 1034 ergs s1) taken from the ATNF catalog ( Manchester et al. 2005) are shown by a dashed blue line. On the right, the distribution of VHE gamma-ray sources as on the left is shown along with the latitude distributions of matter in our Galaxy. The distribution of molecular hydrogen ( H ii; Paladini et al. 2003) is shown by a dot-dashed red line, the distribution of atomic hydrogen ( H i; Dickey & Lockman 1990) is shown by a dotted purple line, and the distribution of CO ( Dame et al. 2001) is shown by a dashed green line.
In such cases the association becomes more difficult to establish. Considering only pulsars with sufficient spin-down luminosity to explain the measured VHE gamma-ray flux, four of the new sources have possible associations with PWNs. However, only in the case of HESS J1825137 is there so far a convincing multiwavelength argument for the offset from the pulsar position in the sense that an X-ray PWN offset in the source direction has been detected. The sensitivity shown in Figure 3 is that for pointlike (